Commits

Branch selector

 d4b2f33

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 11, 2024

• Skip test_fpx_weight_only in fbcode (#1056)

  jainapurva

  authored
  d4b2f33

  Copy full SHA for d4b2f33

• Kleidi 4b blockwise gemv prototype

  Show description for db72dd1
  digantdesai

  authored
  db72dd1

  Copy full SHA for db72dd1

• [Tp Test] Fixe the placment of the device tensor (#1054)

  drisspg

  authored
  5277507

  Copy full SHA for 5277507

• shrink hadamard matrices (#1051)

  Show description for ec860a1
  HDCharles

  authored
  ec860a1

  Copy full SHA for ec860a1

Commits on Oct 10, 2024

• Rename AQT#2 LayoutType -> Layout (#1049)

  jainapurva

  authored
  7038f8b

  Copy full SHA for 7038f8b

• Enable 6-bit kernel

  Show description for 10601b3
  metascroy

  authored
  10601b3

  Copy full SHA for 10601b3

• Introduce lowbit quantized linear MPS kernels

  Show description for ad184da
  manuelcandales

  authored
  ad184da

  Copy full SHA for ad184da

• SpinQuant (#983)

  Show description for 590f8fb
  tobiasvanderwerff

  authored
  590f8fb

  Copy full SHA for 590f8fb

• Change torchao quantization types from int to size_t and preface vars with "preferred_"

  Show description for 76b6e36
  keyan

  authored
  76b6e36

  Copy full SHA for 76b6e36

• Move and rename GranularityType -> Granularity (#1038)

  Show description for 0f6bae5
  andrewor14

  authored
  0f6bae5

  Copy full SHA for 0f6bae5

• Enable AWQ example on CPU (#1043)

  Xia-Weiwen

  authored
  107e378

  Copy full SHA for 107e378

• Aqt rename#1 Layout -> TensorImpl (#1046)

  jainapurva

  authored
  fa6d156

  Copy full SHA for fa6d156

Commits on Oct 9, 2024

• bump version to 0.7 (#1045)

  drisspg

  authored
  a924e6b

  Copy full SHA for a924e6b

• Make module swap the main QAT flow again (#1037)

  Show description for 9e0a59f
  andrewor14

  authored
  9e0a59f

  Copy full SHA for 9e0a59f

• Allow deprecated declarations what using Parallel ExecuTorch

  Show description for 49b1fb6
  Jack-Khuu

  authored
  49b1fb6

  Copy full SHA for 49b1fb6

• Update README.md (#1036)

  jerryzh168

  authored
  900f9ac

  Copy full SHA for 900f9ac

Commits on Oct 8, 2024

• Basic SAM2 AutomaticMaskGeneration example server (#1039)

  cpuhrsch

  authored
  b67407a

  Copy full SHA for b67407a

• Revert "Rename Layout -> TensorImpl" (#1040)

  Show description for 745085f
  msaroufim

  authored
  745085f

  Copy full SHA for 745085f

• Rename Layout -> TensorImpl (#1028)

  jainapurva

  authored
  cc8bf85

  Copy full SHA for cc8bf85

• fix failing fbcode tests (#1033)

  Show description for 35ea27b
  jcaip

  authored
  35ea27b

  Copy full SHA for 35ea27b

• Optimize 3-bit packing

  Show description for 93ff876
  metascroy

  authored
  93ff876

  Copy full SHA for 93ff876

Commits on Oct 7, 2024

• float8 training axiswise scaling support with per-gemm-argument configuration (#940)

  Show description for dec0313
  vkuzo

  authored
  dec0313

  Copy full SHA for dec0313

• add axiswise scaling to Float8Linear (#920)

  Show description for e76db70
  vkuzo

  authored
  e76db70

  Copy full SHA for e76db70

• Activation Aware Weight Quantization (AWQ) (#743)

  Show description for f81fe11
  vayuda

  authored
  f81fe11

  Copy full SHA for f81fe11

• Dynamic Float8 benchmarking llama (#1017)

  jainapurva

  authored
  92dd5f5

  Copy full SHA for 92dd5f5

• add axiswise granularity to Float8Tensor (#919)

  Show description for 52d27a1
  vkuzo

  authored
  52d27a1

  Copy full SHA for 52d27a1

• Refactor `tiktoken` import bare except (#1024)

  Show description for e7331ab
  

  Matthew Hoffman

  authored
  e7331ab

  Copy full SHA for e7331ab

• Use `importlib.util.find_spec` to check if `lm_eval` is installed instead of trying to import it (#1023)

  Show description for e2301e9
  

  Matthew Hoffman

  authored
  e2301e9

  Copy full SHA for e2301e9

Commits on Oct 6, 2024

• [low-bit optim] Fix load state dict when device is different (#1021)

  Show description for c187f87
  gau-nernst

  authored
  c187f87

  Copy full SHA for c187f87

Commits on Oct 4, 2024

• Add Float8 support for AQT tensor parallel (#1003)

  jainapurva

  authored
  9e2a253

  Copy full SHA for 9e2a253

• training ir torchao migration

  Show description for d2982d5
  yushangdi

  authored
  d2982d5

  Copy full SHA for d2982d5

• Unskip `test_choose_qparams_token_asym` in 2.6 (#1004)

  Show description for 5dd0132
  jerryzh168

  authored
  5dd0132

  Copy full SHA for 5dd0132

• Add more videos (#1011)

  msaroufim

  authored
  0cb91ea

  Copy full SHA for 0cb91ea

• Pin torch nightly - Oct 1 (#1008)

  Show description for 0ffbf85
  jainapurva

  authored
  0ffbf85

  Copy full SHA for 0ffbf85

Commits on Oct 3, 2024

• Introduce 6-bit quantization for Llama in torchchat

  Show description for 9ce7ebb
  ramreddymounica

  authored
  9ce7ebb

  Copy full SHA for 9ce7ebb

Pagination

Previous

Next